This study measured incidental vocabulary learning and supported the reliability of Hulstijn and Laufer's (2001) Involvement Load Hypothesis (ILH) among learners with different word levels by focusing on four tasks involving reading the BBC news. A vocabulary test designed by Nation (1983) was taken by the 180 participants, who were then placed into four word levels based on their test results: 40 learners at the 2,000 word level; 60 learners at the 3,000 word level; 48 learners at the 5,000 word level; and, 32 learners at the university word level. Participants at each level were divided into four groups, with each subgroup completing one of four vocabulary learning tasks that varied in the cognitive load required: reading comprehension (low effort), reading comprehension plus supplied target words (moderate effort), reading comprehension plus composition writing (strong effort), and reading comprehension plus dictionary look-up and composition writing (very strong effort). Findings revealed that, in line with the predictions of ILH, vocabulary learning was highest in the fourth task, and descended according to involvement load. This study also suggested that the applicability of incidental learning and ILH was affected by the participants' word levels, and that a critical lexical threshold appeared to exist for comprehending the BBC News.